Machine Learning for Global Precipitation Measurement (GPM) Dual-Frequency Precipitation Radar (DPR) Resolution Improvement
Abstract
The National Aeronautics and Space Administration (NASA) Global Precipitation Mission (GPM) is providing researchers and forecasters with critical new precipitation data from instruments on an international constellation of satellites. Researchers at the University of Alabama in Huntsville Information Technology and Systems Center (UAH/ITSC) and the Universities Space Research Association (USRA) are using use deep learning technologies to downscale (increase) the spatial resolution of the GPM Dual-frequency Precipitation Radar (DPR) data. This process can be used to improve rainfall retrieval estimates from GPM data in areas where ground-based scanning radar data and reliable precipitation gauge observations are lacking. Well maintained gauge networks are sparse, and although ground-based radar is widely available in areas like the United States and Europe, much of the Earth's land mass and almost all of the oceans lack coverage. Improving the resolution of satellite-based radar will thus advance precipitation measurements from space and improve methods of developing GPM precipitation products. The focus of our research is developing a deep learning application that uses convolutional neural networks (CNNs) to learn features that can infer high-resolution information from low-resolution variables. Our research uses coincident low- and high-resolution rain rate gridded data to train a specialized CNN where the low-resolution GPM data is used as input and the corresponding higher resolution ground radar derived data is the target output. The CNN automatically extracts hierarchical features and learns the mappings between sets of corresponding lows- and high-resolution data pairs. The trained CNN can then be effectively used for super-resolution improvement of the low resolution GPM data when no higher resolution ground radar data are available. This presentation will discuss the complexity of downscaling radar data.
- Publication:
-
AGU Fall Meeting Abstracts
- Pub Date:
- December 2018
- Bibcode:
- 2018AGUFM.T31E0363W
- Keywords:
-
- 3399 General or miscellaneous;
- ATMOSPHERIC PROCESSESDE: 1699 General or miscellaneous;
- GLOBAL CHANGEDE: 3099 General or miscellaneous;
- MARINE GEOLOGY AND GEOPHYSICSDE: 8099 General or miscellaneous;
- STRUCTURAL GEOLOGY